2  Prerequisites

Supervised and Unsupervised Text Scaling

2.1 Install OLLAMA

— OLLAMA (Module 4) — To run the code, you must install OLLAMA separately: https://ollama.com/download

2.2 Install R packages

— R packages (all modules) — The code in the lecture relies on a number of R packages. The code also sets up a Python installation through reticulate (this will be used for BERT models). You can install them all at once with the following code:

# ============================================================

# --- R packages ---
pkgs <- c(
  # Text analysis
  "quanteda",
  "quanteda.textmodels",
  "quanteda.textplots",
  "quanteda.textstats",
  "LSX",
  "rollama",
  # Data wrangling & visualisation
  "tidyverse",
  "ggplot2",
  "scales",
  "ggrepel",
  "viridis",
  "corrplot",
  "kableExtra",
  # Classification metrics (Module 4)
  "yardstick",
  # HTTP / API calls (for OLLAMA)
  "httr2",
  "jsonlite",
  "glue",
  # Python bridge (for BERT models)
  "reticulate",
  "corrplot"
)

install.packages(pkgs, repos = "https://cloud.r-project.org")

# --- Python environment (for BERT modules) ---
library(reticulate)

# Create a dedicated virtual environment
virtualenv_create("r-reticulate")
use_virtualenv("r-reticulate", required = TRUE)

# Install Python packages
py_install(
  c("pandas", "openpyxl", "transformers", "torch", "numpy", "scipy", "seaborn", "matplotlib", "python-docx", "nltk"),
  envname  = "r-reticulate",
  pip      = TRUE
)